Hey @leofabri. Great work with the blueprint, it works perfectly with my washer and my dryer! For my dishwasher, there is a long period of time where there is almost no activity, causing a premature ājob completeā state unless I add a really long delay (30+ mins) to capture the last āblipāā¦ This of course will cause the state change to occur much later than desired. Iāve attached what the cycle looks like. Wondering if you have any ideas, or perhaps something that could be done in v3.0 to address this ācycleā. One idea is a setting for āminimum cycle timeā, but ideally there is a less hacky solution.
š Detect and monitor the state of an appliance based on its power consumption - V2.1.1 - Updated!
2 Likes
This blueprint seems nice, though maybe a bit overkill for my situation. I do like the state-machine setup, but I donāt really need the overload or unplugged stuff. Could you maybe make something like a āliteā version with basically just idle, job_ongoing, job_finished or something along those lines?
2 Likes
Hi @Merlijn thank you. Well, in V3.0 Iām planning to add some checkboxes to enable/disable some of the features (Iām testing one for the overload state).
If you donāt want these states though, I could think about a āliteā version that does remove the unused logic. Bear in mind that having the full version or the stripped one likely wonāt affect the overall speed of the automation.
The biggest issue on my side is that Iād have to maintain two versions, and unfortunately, I plan to have a difficult time with that, because this is the one used at home.
Hmm, I see. I never thought that an appliance could spend that long doing nothing!
I donāt know if I understand the āminimum cycle timeā idea. It would filter out the cycles that take less than a minimum time, wouldnāt it?
Based on your chart, the cycle would end at about 8:35 PM, thus ignoring the spike completely, but this would also cause the start to be shifted forward in the future.
Iāve been testing the first unpublished release of V3 that comes with a very similar idea (PM me if you want to be a tester 
). V3 includes an optional āDelayed Job Ongoing timerā that does what I described above.
I know that the most accurate and ideal thing to do is to be able to capture the entire cycle without delays, but I guess that we donāt have enough information to do that.
Updates about V3
Iāve been developing the new version over the past few weeks and have done most of the work. This month, however, Iām very busy with my exams and that means that I postponed the release a bit. However, I still have time to test the changes, and that is positive.
Home Assistant doesnāt seem to have a proper way to notify the users about blueprintsā updates. I donāt want to push too many updates for this blueprint because I know many people wouldnāt update it.
What I prefer to do instead is to release something that I tested, hopefully, bug-free and complete feature-wise.
Feature requests
Please feel free to ask if thereās anything that youād like to see in the next version of the blueprint. I wonāt be implementing everything, but It would give me a better idea of your needs.
1 Like
Thanks for your thoughts!
To elaborate on the āmin cycle time ideaāā¦ The thinking was that a job would go from āongoingā to ācompleteā IF and ONLY IF a āminimum cycle durationā has passed since the start.
For example, my dishwasher āestimated run timeā at the start of cycle is 156 minutes. It may finished a bit sooner/later (maybe +/- 15 minutes), but it would never finish in, for example, 110 minutes. Thus, I would prefer that job_completed only trigger after a min amount of time has passed. For example, I would want the blueprint to assume the job is ongoing while that min time has not been met (and of course while there is power utilization above the finishing threshold).
Not sure this would be useful to others, but that was at least the idea that I had to make this blueprint work for my dishwasher! 
1 Like
This looks really interesting to me.
I see from the requirements the devices need to be plugged into smart sockets that HA can control but also monitor energy usage.
Can anyone recommend decent smart plugs for the uk market that satisfy these requirements?
Thanks
These are what Iām using:
1 Like
I have been trying a number of smart plugs from different vendors, some of which are very complex to setup and maintain in HA.
These ESPHome plugs have opensource firmware and designed to be interfered and maintained though HA so are now my go-to plugs. The setup is the simplest I have seen.
1 Like
The automation keeps firing every 10 seconds for me. It obviously fails on some conditions and nothing happens but surely itās not meant to be triggering ever to seconds?
I think it may be due to very slight power fluctuations in the power monitoring sensor that is changing its state and this firing?
Anything I can do to stop this behaviour?
EDIT: I now see that this is by design - will exclude from recorder/logbook as solution
Hi! Yes, thatās expected. The automation triggers every single time a new power read comes in. Donāt worry about that.
Yes, excluding that from the recorder is a good idea. I will add that to the documentation for the next release!
@Sygnus83 I see. This strange behavior complicates things 
By the way, this doesnāt make any sense! What is the appliance doing from 10:40 to 12:00? It looks like not much is happening. I wonder why the manufacturer wouldnāt just consider the job done much sooner!
The problem is that if the appliance utilizes such a low amount of power, itās difficult to tell if itās still on or not with just the information we have.
Here are a few desperate ideas:
-
I could count the number of spikes, and then āteachā the automation to wait and recognize them. You would need to tell it how many of those you have.
CONS: it wouldnāt be too reliable, and the automation would turn substantially more complex. -
We could implement an external sensor that inhibits the finished state when, for example, an LED or an indicator on the appliance is detected as on. That would be ideal!
CONS: itād require additional hardwareā¦ -
As @dimatx suggested, we could use something like a minimum job_ongoing timer. Then youād set it from the timespan leveraging from circa 11:55 to 8:55, so thatās approximately 3 hours. Immediately after that, the automation would behave like normal and use the Delayed Job Completion timer for the last part, as it usually would.
CONS: what if your appliance supports more than one program with different durations? This method wouldnāt account for that.
This is probably the best idea, so far. -
Anything better?
I wish I could make something that works for everybody, but you have to consider that we have some constraints. And conceptually, automations are something that is usually very repetitive and stateless; so you can forget any advanced power analysis.
Hi !
Thanks for the answer, unfortunately I donāt know what the best solution would be either.
I think this time is ādryingā. The pump switches on twice and lets in water after the first one. In this case, the water from the hot plates splashes onto the cold inner wall.
Development Updates
Hey folks, Iām back on the project!
Today, V3.0.0-dev branch came to GitHub. Even if itās not ready for a release, you can already see what changed. I know itās taking some time, but Iām trying to make this repo as streamlined as possible.
The automation itself seems very stable so far (Iāve been testing it for over a month now), but there is a lot of documentation to review/write as there are important changes.
Speaking about these ideas: Iām still evaluating if these are going to get implemented, or if Iām just going to release what I did so far.
2 Likes
I think @Sygnus83 is right. For efficiency reasons, some dishwashers use residual heat to dry the dishes, rather than turning on heaters. I think that time is used to dry the dishes, so itās doing stuff, just using electricity infrequently and in a limited capacity. Sygnus - Is yours also a Bosch?
Hi @dimatx, my dishwasher does that as well. However, I can tell itās doing something because the power consumption is still higher compared to the idle state (~ 0.6 Watts).
Take a look at my dishwasher (Miele):
This was done with grafana, just to highlight the min power usage while in operation.
At this point, I have two suspects:
- Your dishwasher is unbelievably efficient, and during the drying process, it doesnāt consume more power than when itās offā¦ itās hard to believe
- Your socket is not sensitive/precise enough. This is the most plausible explanation. If this is proven to be the case, the only solution I can think of is to try a different smart plug.
Thatās fair, @leofabri Itās a Tasmota-flashed plug, so I perhaps I can up the sensitivity of said power monitoring. Perhaps it could be re-configured.
PS. Just curious, do you pull directly into Grafana from the HA db, or do you feed it into InfluxDB and then have grafana pull from there?
Well, maybe! I think I canāt help with that though. Iām using a different smart socket (I was just mentioning it here in V3.0.0-Dev).
Well, Iām using Grafana with InfluxDB. You know, Iām a data person and the sole idea of losing it scares me
1 Like
OK, since you like data ā¦ Hereās some data on the end of the last cycle. I increased the power monitoring precision to include a decimal point, so it no longer rounds to the nearest Watt.
(1) End of the cycle, per dishwasher
(2) 0.9W as measured by power monitoring
(3) 0.8W as measured by power monitoring
(4) 0.7W as measured by power monitoring
I will have to try to replicate this across multiple runs, but seems like I may be able to use ābelow 0.8Wā as my completion condition. I will test it and report back.
